Whole Power Server + Virtual Server Monitoring - Part 5 via Ganglia
nagger 100000MRSJ Comment (1) Visits (23768)
Ganglia is a open source performance and configuration tool that collects data from a tiny daemon on each OS and then creates a Web server based graphical tools to draw performance data and show the configuration. With a few add-ons, you can dynamically decide the historical data you want graphed and with POWER add-ons you get the PowerVM stats for shared processors, Entitlement, physical CPU use etc. Under the covers it is using the amazing rrdtool so you can extract the data for other uses and it automatically space manages the data.
By using the cluster feature you also get the CEC view of a whole machine and what the Virtual Machines (LPARs) are doing. You can also via a simple ganglia command collect other data on a VM of any type, so, for example, the number of users online or transaction rates of a RDBMS and the really cool thing is Ganglia will automatically start graphing the data after a few minutes. I am rarely impressed these days but Ganglia is amazing. One of those "gosh I wish I had written that myself". When asked when is nmon going to collect data from many VM/LPARs the answer is "take a look at Ganglia" it is way better than anything I could have done.
Here are some sample graphs to give you the idea:
Regular user interface for Ganglia showing a summary of a whole machine:
With the Power add-on we call a cluster all the Virtual Machines (LPARs) of the machine a cluster and get a whole machine view for day
- this machine has 2 CPUs and the workload is using them all most of the time:
There are lots more that are graphed at the whole machine level like total disk read and write plus network transmit and receive here is the run queue graph:
For more information take a look at: